Chunked parquet dataset (~1.3 GB total, 1,686,039 rows).
The original single parquet was split into 17 chunks of ~78 MB each so the
data can ship as plain git files (no Git LFS needed; every chunk is well
below the 100 MB GitHub per-file limit).
Files
File
Rows
Size
part_000.parquet ... part_015.parquet
100,000 each
~78 MB each